RNA Gainof Function Model for Myotonic Dystrophy

Myotonic Dystrophy Associated Microsatellite Expansions Are Toxic at the RNA Level

Although unstable microsatellite expansions can influence gene expression at multiple levels, several observations suggest that adult-onset DM is an RNA gain-of-function disease. The first clue that mutant DM RNA molecules are unusual was based on computer modeling which predicts that these d(CTG) expansions form stable double-stranded (ds) RNA structures or RNA hairpins (Zuker et al. 1999). Indeed, the existence of these RNA structures was demonstrated using chemical and enzymatic structure probing, thermal de-naturation, magic angle spinning solid-state NMR and visualization of rodlike RNA duplexes in the electron microscope (Napierala and Krysysosiak

1997; Michalowski et al. 1999; Tian et al. 2000; Sobczak et al. 2003; Leppert et al. 2004). The r(CCUG) repeats in mutant ZNF9 transcripts also form RNA hairpin structures (Sobczak et al. 2003). However, there are significant differences in the stability of ds r(CCUG) versus ds r(CUG) with tandem C : U and U : C mismatches and a larger terminal loop for ds r(CCUG) hairpins compared with U : U mismatches for ds r(CUG).

Although DM repeat expansions affect both DNA and RNA structures, another distinguishing attribute of DMPK and ZNF9 mutant allele transcripts is that they are retained in nuclear foci while normal transcripts are exported to the cytoplasm. These ribonuclear foci, which were originally detected by RNA fluorescence in situ hybridization (FISH) analysis, do not colocalize with any known nuclear structures, including splicing factor compartments or speckles, Cajal bodies, the perinucleolar compartment and promyelocytic leukemia nuclear bodies (Taneja et al. 1995; Davis et al. 1997). RNA FISH also indicates that DM2 ribonuclear foci in skeletal muscle are more intense and larger than DM1 ribonuclear foci, perhaps reflecting higher ZNF9 expression levels (Mankodi et al. 2003). The discovery of these novel nuclear structures was particularly striking in light of concurrent studies on coding-region microsatellite expansion diseases, such as Huntington's disease (HD) and the spinocerebellar ataxias (SCAs). In HD and the SCAs, d(CAG)„ expansions result in the synthesis of proteins containing a toxic polyglutamine (polyQ) region which accumulates in intranuclear inclusions (reviewed in Landles and Bates 2004; Taroni and DiDonato 2004).

Do the DM1 and DM2 expansion mutations affect the processing of their host transcripts? In contrast to an earlier report, mutant DMPK transcripts are correctly spliced and polyadenylated (Wang et al. 1995; Davis et al. 1997). FISH analysis indicates that these mRNA molecules remain intact even within RNA foci since hybridization signals using probes against the first seven DMPK exons and the d(CTG)n repeat colocalize to these foci (Taneja et al. 1995). This result is in agreement with the majority of expression studies which have reported only modest changes in DMPK RNA levels while DMPK protein levels decline (reviewed in Nykamp and Swanson 2004). In contrast, recent work suggests that ZNF9 RNA and protein levels are unaffected in DM2 heterozygous and homozygous individuals (L. Ranum, personal communication). The processing of ZNF9 pre-mRNA is probably not adversely influenced by the DM2 expansion mutation because it is positioned in the first intron approximately 850 nucleotides upstream of the 3' splice site of ZNF9 exon 2.

Although these observations suggested that poly r(CUG) and poly r(CCUG) exist as dsRNAs which accumulate in ribonuclear foci, transgenic mouse studies were required to confirm RNA-mediated pathogenesis as a viable disease model (reviewed in Wansink and Wieringa 2003). Transgenic mice carrying a greater than 45 kb fragment from the DM1 locus, which contains the DMWD, DMPK and SIX5 genes as well as a DMPK d(CTG)3oo expansion, develop myotonia and DM-associated muscle histopathology (Seznec et al. 2001). Interestingly, other effects of transgene expression that are unrelated to DM disease, such as elongated crossed teeth, were also present. The possibility that d(CTG)n expansions alone are toxic independent of gene context was tested by creating mouse lines carrying a human skeletal actin (HSA) transgene with either a d(CTG)5 (HSAsr) or a d(CTG)250 (HSALR) repeat tract inserted into the HSA 3'-UTR (Mankodi et al. 2000). While the HSAsr mice are indistinguishable from normal sibs, HSALR mice develop skeletal muscle myotonia, centralized myonuclei and split myofibers characteristic of DM disease. Notably, several lines were created which express different levels of the transgene and HSALR mice with no, or relatively low levels of, transgene expression are not affected by myotonia, while high expressers develop robust myotonia. This result, together with the discovery that DM1 and DM2 are caused by structurally related repeat expansions in unrelated and unlinked genes, provides strong support for the conclusion that DM is an RNA gain-of-function disease which results from the expression of pathogenic ds r(CUG) and ds r(CCUG) RNA molecules.

Toxic RNAs Molecules Sequester Muscleblind-like Proteins

While the minimal microsatellite expansions associated with disease vary between DM1 and DM2, the predicted stability of the respective dsRNAs is remarkably similar [approximately 70 kcal/mol for both r(CUG)5o and r(CCUG)75]. Why are poly r(CUG) and poly r(CCUG) RNA molecules toxic above a certain repeat length? One possibility is that these RNA molecules are high-affinity binding sites for cellular factors. Binding of these factors might be proportional to the number of repeats and thus they are effectively sequestered above a certain length threshold. As the name implies, CUGBP1 was the first r(CUG)-binding protein identified and it is the founding member of the mammalian CELF family of RNA-binding proteins that contain three RNA recognition motifs (Caskey et al. 1996; Timchenko et al. 1996; Good et al. 2000; Ladd et al. 2001). However, several properties of this protein make it an unlikely candidate for a sequestered factor in DM. While disease-associated r(CUG) and r(CCUG) repeats form RNA hairpins, CUGBP1 is a single-stranded (ss) RNA-binding protein that recognizes r(CUG) trinu-cleotide and UG dinucleotide repeats (Timchenko et al. 1996; Michalowski et al. 1999; Takahashi et al. 2000). Although r(CUG)8 binding activity and protein levels increase in DM1 cells and skeletal muscle, CUGBP1 does not colocalize with ribonuclear foci. Thus, CUGBP1 activity appears to be indirectly influenced by poly r(CUG) and poly r(CCUG) expression (Timchenko et al. 1996; Savkur et al. 2001; Ho et al. 2004).

In contrast to CUGBP1, considerable evidence now suggests that the muscleblind-like (MBNL) proteins are the sequestered factors in DM (Fig. 1a).

DMPK mRNA ZNF9 pre-mRNA

Gene Splicing Pattern Phenotype wt adult clcni n—n—H & I my°tonia

DM/wt neonate wt adult

DM/wt neonate wt adult

TNNT2 B—0—0 cardiomyopathy

DM/wt neonate wt adult

DM/wt neonate wt adult

MAPT

wt adult

DM/wt neonate wt adult

SERCA2

DM/wt neonate

^ TNNT3

Fig. 1 Muscleblind-like (MBNL) loss-of-function model for myotonic dystrophy (DM). a Expression of mutant DMPK messenger RNA [mRNA; coding region, black box; 3' and 5' untranslated regions (UTRs), line; poly(A) tail, (A)n] or ZNF9 precursor mRNA (pre-mRNA, exons, black boxes; 3'- and 5'-UTRs, open boxes; introns, lines) leads to sequestration of the MBNL proteins (ovals) on double-stranded (ds) r(CUG) and ds r(CCUG) RNAs, respectively. The arrows indicate that the affinities of the MBNL proteins for ds r(CUG) and ds r(CCUG) are relatively high. b Loss of MBNL, or upregula-tion of CELF, proteins leads to retention of neonatal isoforms (exons, numbered black boxes; introns, horizontal lines; splicing pattern, angled lines) in adult tissues, which, in turn, results in distinct pathophysiological effects (e.g., myotonia). The connection between TNNT2 missplicing and DM-associated heart defects (cardiomyopathy, conduction block) has not been established. c Missplicing of APP, GRIN1 (NMDA R1), MAPT, RyR1, SERCA1, SERCA2 and TNNT3 in DM tissues results in neonatal exon retention in adults but the phenotypic effects have not been determined

The MBNL proteins were originally identified on the basis of their unusual ability to bind and photo-cross-link to ds r(CUG), but not to ss r(CUG) or other ds RNA molecules [ds r(CAG), HIV TAR ds RNA], in HeLa nuclear extracts (Miller et al. 2000). MBNL binding to ds r(CUG) is proportional to repeat length in vitro and these proteins colocalize with poly r(CUG) and poly r(CCUG) RNA foci in cotransfected cells as well as DM skeletal muscle and cortical neurons (Fardaei et al. 2001, 2002; Mankodi et al. 2001, 2003; Jiang et al. 2004). While several additional nuclear RNA-binding proteins, such as hnRNPs F and H, also accumulate in ribonuclear foci to a much lesser degree, other ss RNA- and ds RNA-binding proteins (2/,5/-OAS, ADAR, CUGBP1, CUGBP2/ETR3, FLAP-1/LRRFIP1, hnRNP A1, hnRNP I, hnRNP M, KSRP, HuR, NF90/ILF3, PACT/RAX, PKR, RNA helicase A) and DNA-binding proteins (Sp1, RARy) do not (Mankodi et al. 2003; Jiang et al. 2004; Kim et al. 2005). The observation that the ss r(CUG)-binding proteins CUGBP1 and CUGBP2/ETR3 do not colocalize with either DM1 or DM2 ribonuclear foci supports previous suggestions that these nuclear structures contain primarily ds r(CUG) and ds r(CCUG). Significantly, RNA FISH combined with immunocytochemistry indicates that formation of ribonuclear foci in DM1 cortical neurons correlates with a decrease in the diffuse nuclear, or nu-cleoplasmic, population (Jiang et al. 2004). Interestingly, three proteasome subunits (20Sa, 11Sa, 11Sy) also colocalize with neuronal ribonuclear foci, suggesting that functional depletion of MBNL might result from targeted protein turnover. Although it is tempting to speculate that the formation of these ribonuclear foci is a primary event in the DM pathogenesis pathway, complexes between MBNL and ds r(CUG) and ds r(CCUG) RNAs which exist outside of these foci might also effectively sequester MBNL proteins (Ho et al. 2005b).

Poly r(CUG) Toxicity Requires Expression of Specific Muscleblind-like Isoforms

Muscleblind proteins were originally identified as factors required for late-stage development of muscle and eye tissues in Drosophila (Begemann et al. 1997; Artero et al. 1998). In humans, there are three MBNL genes (MBNL1, MBNL2, MBNL3) (Miller et al. 2000; Fardaei et al. 2002; Squillace et al. 2002). While MBNL1 and MBNL2 are expressed in a variety of tissues, MBNL1 mRNA levels are high in heart and skeletal muscle. MBNL3 expression appears to be restricted to only a few tissues, including placenta. All three MBNL proteins colocalize with r(CUG) repeats in cells cotransfected with d(CTG) repeat and green fluorescent protein (GFP)-MBNL expression plas-mids (Fardaei et al. 2002; Ho et al. 2005). Moreover, MBNL1 and MBNL2 accumulate in ribonuclear foci in neurons (Jiang et al. 2004). Intriguingly, MBNL1 proteins may play a fundamental role in ribonuclear foci formation and/or maintenance because small interfering RNA (siRNA)-mediated knockdown of MBNL1 mRNA leads to a substantial loss (approximately 70%) of these foci (Dansithong et al. 2005). Similar reductions of CUGBP1 and MBNL2 mRNA levels resulted in a smaller effect (approximately 20%) on the number of ribonuclear foci.

Are ds r(CUG) and ds r(CCUG) RNA molecules, or ribonuclear foci, inherently toxic to metazoan cells or does toxicity result from MBNL protein sequestration? In support of the latter possibility, a recent study suggests that ds r(CUG) RNA chains are not toxic in Drosophila (Houseley et al. 2005). Transgenic flies expressing GFP-DMPK-(CTG)n_162 3'-UTR fusions develop ribonuclear foci in some larval and adult muscles, but not in neurons, only when d(CTG)162 is expressed. Although endogenous muscleblind proteins colocalize with these foci, these flies are viable, overtly normal and have extended lifespans. Interestingly, fly muscleblind proteins are not required for foci formation but coexpression of human MBNL1 results in the appearance of neuronal ribonuclear foci. It is important to note that the Drosophila muscleblind proteins vary from the vertebrate MBNL homologues since they possess only two of the four CCCH (C3H) zinc-finger-related motifs which are required for high-affinity ds r(CUG) binding in vitro (Fig. 1) (Miller et al. 2000; Yuan et al., unpublished data). These results support the hypothesis that pathogenesis associated with d(CTG)n expression is mediated by interactions with specific MBNL proteins which are expressed in vertebrate cells.

If DM disease results from loss of certain MBNL isoforms owing to sequestration by toxic poly r(CUG) and poly r(CCUG) RNA molecules, then disease-associated phenotypes common to DM1 and DM2 should be recapitulated in Mbnl knockout mice. This possibility has been tested by generating mice which fail to express the 40-43-kDa isoforms which utilize an initiation codon in exon 3 of the Mbnl1 gene. These larger isoforms bind, and photo-cross-link, to ds r(CUG), while the shorter 26-36-kDa Mbnl1 proteins do not. Mice carrying a homozygous Mbnl1 exon 3 deletion (Mbnl1AE3/AE3) are viable but develop the most characteristic features of adult-onset DM, including myotonia, dustlike cataracts and heart conduction defects (Kana-dia et al. 2003a, and unpublished data). Since the adult-onset disease can be modeled in Mbnl1 knockout mice in the absence of toxic ds r(CUG) and ds r(CCUG) RNA and ribonuclear foci, the striking conclusion is that DM is an MBNL loss-of-function disease resulting from an RNA gain-of-function mutation.

The Splicing Connection: DM is Associated with Fetal Exon Retention in Adults

Mutations associated with a large number of inherited diseases result in the perturbation of normal patterns of pre-mRNA splicing (Faustino and Cooper 2003; Garcia-Blanco et al. 2004; Matlin et al. 2005). In humans, multiexon genes are generally alternatively spliced. During postnatal development, fetal tissues are remodeled by a series of alternative splicing events to generate specific isoform ratios. Generally, these splicing decisions are temporally coordinated during the postnatal period so that tissues at various stages of maturation are responsive to the specific physiological demands characteristic of each developmental interval. How are these splicing decisions regulated so precisely so that the correct protein isoforms are synthesized at the proper time? Surprisingly, studies designed to reveal DM pathogenesis have provided fundamental insights into the regulation of pre-mRNA alternative splicing during the fetal-to-adult transition period.

In DM1 and DM2, the processing of DMPK and ZNF9 mutant allele transcripts does not appear to be significantly affected by r(CUG)n and r(CCUG)n expansions; however, the alternative splicing of other transcripts is influenced. The discovery of misregulated splicing in DM1 resulted from studies designed to define the RNA sequence elements, and the corresponding transacting binding partners, which regulated the alternative splicing of exon 5 of chicken cardiac troponin T (TNNT2/cTNT) (Phillips et al. 1998). The splicing of TNNT2 exon 5 is developmentally regulated with inclusion favored in embryonic tissues, while skipping of this exon is the predominant pattern in adults. For chicken cTNT, exon 5 splicing is regulated by four muscle-specific splicing enhancers (MSE1-MSE4) in intron 5 and MSE1 and MSE4 each contain r(CUG)2 repeats. Human cTNT contains an r[CUG(N)9(CUG)2C(CUG)2] repeat motif and exon 5 inclusion is favored in DM1, but not in normal, adult heart muscle (Fig. 1b) (Phillips et al. 1998; Ladd et al. 2001). These results suggest that retention of TNNT2 exon 5 in adults might contribute to the heart conduction defects characteristically seen in DM. However, the connection between aberrant TNNT2 RNA splicing and DM disease remains tenuous. In DM1 and DM2, progressive cardiac conduction impairment, including atri-oventricular (A-V) block, atrial fibrillation and ventricular/supraventricular arrhythmias are the most common heart problems (Pelargonio et al. 2002; Schoser et al. 2004b). Dilated cardiomyopathy has been documented for several DM2 patients and cardiomyopathies also develop in some DM1 patients. Interestingly, TNNT2 mutations are generally linked to hypertrophic car-diomyopathy and dilated cardiomyopathy (DCM) and abnormal cTNT pre-mRNA splicing also occurs in mammals prone to DCM (Watkins et al. 1995; Biesiadecki et al. 2002; Pelargonio et al. 2002; Schoser et al. 2004b). Thus, TNNT2 pre-mRNA missplicing may be one component of the heart conduction defect common to DM1 but additional contributing factors will probably be uncovered in the future. Splicing of myotubularin-related 1 (MTMR1) is also dysregulated in DM1 adult heart with enhanced retention of the fetal A isoform (Ho et al. 2005a).

A more convincing argument for a direct role of disrupted RNA splicing in DM pathogenesis is provided by the myotonia which is a characteristic feature of DM1 and DM2. Mutations in both the skeletal muscle sodium (SCN5A)

and chloride (CLCN1/C1C-1) channels cause myotonia in humans (Chen et al. 1997; Pusch 2002). DM-relevant myotonia has been linked to a defect in the CLCN1 channel but in this disease CLCN1 missplicing is the underlying pathogenic event (Fig. 1b) (Charlet-B et al. 2002; Mankodi et al. 2002). In normal adults, CLCN1 exons 6, 7 and 8 are spliced together directly to generate functional chloride channels. During the fetal and neonatal periods and in either HSALR or Mbnl1AE3/AE3 adult knockout mice, intron 2 and exons 6b, 7a and 8a are frequently included (Mankodi et al. 2002). These intronic and ex-onic sequences contain in-frame termination codons which make the resulting mRNA susceptible to turnover by the nonsense-mediated decay (NMD) pathway. For mRNAs that escape NMD, translation of truncated CLCN1 proteins has a dominant-negative effect on chloride channel function (Berg et al. 2004). Additionally, missplicing of skeletal muscle TNNT3 pre-mRNA has been documented in DM1 as well as in the HSALR and Mbnl1AE3/AE3 mouse models but the physiological effects, if any, of adult expression of fetal TNNT3 isoforms are unknown (Kanadia et al. 2003b). Missplicing of the skeletal muscle ryanodine receptor RyR1 and sarcoplasmic/endoplasmic reticulum Ca2+-ATPase (SERCA) 1 and 2 has also been reported and may account for altered calcium homeostasis in DM myotubes (Fig. 1c) (Kimura et al. 2005). The splicing of MTMR1 pre-mRNA is also abnormal in DM1 skeletal muscle (Buj-Bello et al. 2002).

Insulin resistance, which is another characteristic pathological feature of DM, is also caused by abnormal developmental regulation of splicing (Savkur et al. 2001). The fetal splicing pattern for the insulin receptor (IR) is inclusion of exon 11, which generates the lower signaling IR-A isoform. While IR exon 11 is included in normal adults, this exon is skipped in DM1 and DM2 adults (Fig. 1b). Thus, current evidence supports the conclusion that retention of the fetal TNNT2, CLCN1 and IR splicing pattern is responsible for the cardiac, myotonia and insulin resistance characteristic of DM disease.

DM has a significant effect on the function of the central nervous system (CNS), with distinctive behavioral effects and hypersomnia in adults as well as mental retardation in the congenital disease. As described previously, DMPK is expressed in the CNS and r(CUG)n expansions accumulate in neuronal nuclei. These CNS defects might result from abnormal splicing of several pre-mRNAs, including amyloid precursor protein (APP), microtubule-associated protein tau (MAPT) and the glutamate receptor, N-methyl-D-aspartate 1 (GRIN1/NMDA R1) (Fig. 1c) (Sergeant et al. 2001; Jiang et al. 2004). In conclusion, expression of the DM1 and DM2 expansion mutations clearly perturbs the regulation of RNA alternative splicing during postnatal development. Nevertheless, the molecular events underlying additional manifestations of DM disease (hypersomnia, mental retardation, muscle weakness/wasting, testicular atrophy, hypogammaglobulinemia, cataracts) have yet to be elucidated.

MBNL and CELF Proteins are Splicing Antagonists Which Regulate Fetal Exon Splicing

How are the expression of r(CUG) and r(CCUG) repeat expansions and sequestration of the MBNL proteins related to aberrant splicing during development? CUGBP1 was the first RNA-binding protein implicated in DM pathogenesis and was initially characterized as an r(CUG)8-binding protein (Tim-chenko et al. 1996). Enhanced r(CUG)-binding activity was observed using extracts prepared from DM1 cells and subsequent analysis demonstrated that CUGBP1 is a splicing factor. Cells cotransfected with cTNT minigene reporter and CUGBP1 protein expression plasmids show enhanced cTNT exon 5 inclusion and this effect is abolished by mutation of the r(CUG) repeats to r(CAG) (Philips et al. 1998). Although CUGBP1 activity and steady-state protein levels are elevated in DM1 muscle and myoblasts, the connection between increased CUGBP1 splicing activity and expression of mutant DM1 and DM2 RNA was obscure until recent results became available that linked CUGBP1 and MBNL splicing activities (Savkur et al. 2001; Timchenko et al. 2001a; Dansithong et al. 2005). Intriguingly, CELF and MBNL protein families are antagonistic regulators of fetal exon splicing (Ho et al. 2004). Cotransfection analysis was used to demonstrate that overexpression of MBNL1, MBNL2 or MBNL3 proteins results in either enhanced skipping of TNNT2 exon 5 or increased inclusion of IR exon, which is identical to the normal adult splicing pattern. Alternatively, the DM splicing pattern is seen following siRNA-mediated knockdown of MBNL1 in HeLa cells, which promotes TNNT2 exon 5 inclusion and IR exon 11 skipping. Mutational analysis revealed that several MBNL1 binding sites (consensus is YGCUU/GY) exist immediately upstream of the 3/ splice site of TNNT2 exon 5 and mutation of these sites abolishes the effect of MBNL overexpression on exon 5 splicing (Ho et al. 2004). The similarity in Tnnt2 and Clcn1 pre-mRNA splicing patterns between Mbnl1AE3/AE3 knockout and CUGBP1 transgenic (MCKCUG-BP1) mice also indicates that the MBNL and CELF protein families are antagonistic splicing regulators in vivo (Kanadia et al. 2003a; Ho et al. 2005a). CELF-MBNL interactions may also function in additional posttranscriptional regulatory pathways since CELF proteins have been implicated in RNA editing as well as mRNA translation and turnover (Anant et al. 2001; Timchenko et al. 2002, 2005; Mukhopadhyay et al. 2003; Iakova et al. 2004; Baldwin et al. 2004).

Is Myotonic Dystrophy Caused by MBNL Loss, CUGBP1 Overexpression or Both?

O the basis of the results described in the preceding sections, the original MBNL loss-of function model proposed for DM pathogenesis can be updated (Miller et al. 2000). Certain genes implicated in tissue-specific effects in DM, includ ing CLCN1 and IR, contain fetal exons and inclusion of these exons during pre-mRNA splicing is promoted by CELF activity. In contrast, adult splicing patterns are triggered by activation of MBNL sometime during the neonatal-to-adult transition. Alternatively, CELF protein activity may decline during this transition as suggested by a recent study which demonstrated that CUGBP1 and CUGBP2/ETR-3 protein levels are relatively high in embryos and low in most adult somatic tissues, with the striking exception of brain (Ladd et al. 2005). Loss of MBNL activity, either by sequestration on ds r(CUG) and ds r(CCUG) RNA molecules in DM tissues and HSALR skeletal muscle, or in all tissues in Mbnl1 knockout mice, leads to fetal exon retention in adult mRNA molecules because CELF splicing activity is unopposed. This simple model is appealing since it accounts for the increase in CELF splicing activity in DM tissues and cells due to loss of the MBNL splicing antagonist. However, this MBNL1 loss-of-function model fails to explain the observed increase in CUGBP1 steady-state tissue levels in DM1 skeletal muscle and myoblasts as well as the elevated r(CUG)8 RNA-binding activity in vitro (Timchemko et al. 1996; Savkur et al. 2001; Dansithong et al. 2005). Unfortunately, these effects on CUGBP1 protein levels may be specific to humans since they are not reproduced in either HSALR or Mbnl1AE3/AE3 mice. Another puzzling observation argues against the MBNL loss-of-function model. Mutation of the CUGBP1 binding site downstream of TNNT2 exon 5 does not affect MBNL1 splicing regulation since siRNA-mediated knockdown of MBNL1 levels still leads to enhanced exon 5 inclusion in transfected HeLa cells (Ho et al. 2004). In contrast to the prediction of the model, this CUGBP1 mutant binding site minigene no longer responds to expression of an r(CUG) repeat expansion RNA; therefore, siRNA-induced depletion of MBNL1 may not be synonymous with loss due to sequestration by r(CUG)n expansions. Nevertheless, this conclusion is tentative because we do not know how MBNL proteins interact with either precursor RNAs or ds r(CUG). The r(CUG)„ expansion (CUG960) used in this study consisted of discontinuous r[(CUG)20CUCGA]48 repeats so the affinity of MBNL proteins for these repeats might be low relative to that for the continuous DM repeats. An alternative conclusion is that the TNNT2 intron 5 mutation creates a higher-affinity binding site target for MBNL binding which effectively competes with CUG960 binding activity. In summary, current evidence suggests that both MBNL and CELF protein activities in RNA splicing, and potentially other posttranscriptional regulatory steps, are adversely affected in DM.

Was this article helpful?

0 0

Post a comment